CentiScape User Guide

CentiScape is a Java plugin for Cytoscape, a software platform for the analysis and visualization of molecular interaction networks. The plugin computes specific centralities parameters describing the network topology. CentiScape is a project of the Center for Biomedical computing (CBMC), University of Verona, Italy.

Feature List

Computed Network Parameters

Average Distance Diameter

Computed Centralities

Degree Stress Betweenness Radiality Closeness Centroid Value Eccentricity Eigenvector Bridging Centrality Edge Betweenness (for edges)

Centralities are computed for directed, undirected and weighted networks. For a formal description of the centralities see the Network Centralities documentation available athttp://www.cbmc.it/~scardonig/centiscape/centiscape.php

How to do

Copy the CentiScape.jar file to your [Cytoscape_Home]/plugins directory and start Cytoscape.

Load the network which you want to apply the plugin. From the Main Menu, Select "PlugIns" --> "CentiScape" .

A new panel will appear in the left side of your Cytoscape as shown in figure 1. The panel shows you the centralities you can select. You can select all the centralities using the 'select all' button. If you want to select some of them from the list click with your mouse on the corresponding button.

figure 1

figure 1

On the right side of each centrality button there is also an help button. If you click on it a brief description of the centralities will appear. In figure 2 is shown the Stress Help windows.

When you have chosen and selected the Centralities you need you have to click on the 'Start' button to start the computation. You can also choose directed or undirected networks, and weighted networks (if you have a value for the edges to be considered as a distance in the computation). The computation may need a long time, depending on the size of your network. A banner and a node worked count appear during the computation.

We have three kind of Parameters:

Network Parameters

Node Parameters

Edge Parameters

Network Parameters

The network parameters concern the entire network and not the single nodes. They are Diameter and Average Distance. If you have selected them they will appear on the data panel selecting the Network attribute browser.

figure 2

Node and edge Parameters

All the other centralities are node parameters and concern the single nodes, except of Edge Betweenness that is an edge parameter (It will appear as an edge attribute). So they will appear on the Data Panel as node attributes. Using the Node attribute browser you can select one or more of them as normal attributes.

CentiScaPe also calculates the min, max and mean value for each centrality. Since they are network parameters they appear on the data panel on the Network attribute browser.

As for the other attributes you can save and load Network and Node parameters to/from a file. If you load an attribute and then you try to recalculate it, a warning message will appear. This is because the computation may need a long time and it is better not to recalculate parameters if you do not have changed the network structure. To use loaded attributes (not only centralities attributes) you have to click on the “Start with loaded attribute” button after loading the values from file. This will not start a new computation but will open a result panel using the attribute values you have loaded from file. Remember that changing the structure of a network the centralities values will change and you need a new computation to obtain the new right values.

CentiScape Results Panel

If you have selected one or more Node centralities, a result panel will appear on the right side of your Cytoscape window (figure 3).

A detail of a CytoScaPe result panel is shown in figure 4.

The CentiScaPe result panel shows you a slider for eachcentrality that has been calculated. The threshold is initialized to themean value of each centrality. The slider threshold is initialized to themean value of each centrality and all the node that have a centralityvalue less or equal to the threshold are highlighted with a yellowcolor in the network view. So if you have selected one centrality allthe node that have a value less or equal the threshold for thatcentralities are colored with yellow. If you have selected more thanone centralities you can decide to combine them with an AND or anOR operator:

figure 3

figure 4

If you select the AND operator the node for which all the value are less or equal the thresholdare highlighted.

If you select the OR operator the node for which at least one value is less or equal thethreshold are highlighted.

figure 4

You can also deactivate one or more centralities if you do not need them, clicking on thecorresponding checkbox. You can also select to highlight the nodes that are more/equal than thethreshold selecting the corresponding button for each centrality. Therefore you can select themore/equal option for some centralities, the less/equal option for others and can join them with theAND or the OR operator. If you need to be more precise you can modify the threshold writing thenumber in the corresponding text pane and then clicking the 'update' button. The 'restore' button willreset the threshold to the mean value. In the figure 5 are highlighted all the nodes that have centroidvalue more/equal than the threshold.

Plot by node option

When you have highlighted the nodes you are interested in, you can also select one of them and tohave a graphical view of all its centrality values. You have to select it in the combo box and to clickon the plot button. In figure 6 is shown the graphical output for the node YBR160W in the networkof figure 5. the Centralities selected are Radiality, Betweenness and Stress.

figure 5

The bar graph shows the mean value (blue), the min value (green), the max value (white) and the node value (red). Because of visualization problem all the values are normalized in a scale from 0 to100. If you select one bar of the graph the exact value will appear.

Plot by centrality option

The plot by centrality visualization is an easy way to discriminate nodes that are relevant in thenetwork for two selected parameters. To use it you have to select the two parameters on the “plot bycentrality menu” and then you have to click on the corresponding “plot” button. It results in a chartwhere there is an attributes on the horizontal axis and another attributes on the vertical axis. Youcan use this option in four different ways:

Two different Centrality attributes A Centrality attribute and a generic double attribute Two generic double attributes One centrality versus itself

figure 6

figure 7

The first three ways are similar. The plot by centrality shows clearly the distribution of the twoselected parameters in the network (in figure 7 you can see the centrality/centrality option). You canselect also one centrality attribute and a generic double attribute as for example the phosphorylationlevel. Another way to use it is to visualize the scatter plot of two generic double attributes. This isan extra function of the plug-in an can be used in the same way of the centrality/centrality optionand centrality/generic attribute option.

• nodes having high values for both attributes are in the top right corner.• nodes having one attributes with an high value and the other with low value are in the

top-left corner or in the bottom-right corner.• nodes having low values for both attributes are in the bottom left corner.• nodes with both values around the mean are in the middle of the graph.

Passing with the mouse on one node in the graph its identifier appears. In the figure there is an example with radiality in the vertical axis and node degree in the horizontal one. The only node having high radiality and high degree is easily identified in the top right corner of the graph.

If you use the plot by centrality option selecting the same centrality for both the horizontal and the vertical axis the result is obviously a straight line. In the left part of the graph you see the nodes

having low value for thecentrality selected. Thevalue grows going to the right part of the graph (see figure 8)

Bugs/Feature Requests

If you encounter a bug with this plugin, or have a feature suggestion, weencourage you to use the Cytoscape Bug Tracker:http://www.cbio.mskcc.org/cytoscape/bugs/.

If you log a bug, we will automatically email you when the bug is resolved.

Updates regarding this plugin will be posted to the Cytoscape-announcemailing list. You can subscribe to the mailing list or browse the archivesat: http://groups-beta.google.com/group/cytoscape-announce.

figure 8

Contacts

CentiSCaPe is a project of the

Center of Biochemical Computing (CBMC) University of Veronawww.cbmc.it

For any questions concerning this plugin, please contact:

Giovanni Scardoni : giovanni.scardoni@gmail.com

This software is made available under the GPL (General Public License).